Windward Intelligence: Only 32 Vessels Transited Hormuz on June 20; 11 Dark Transits and 193 GPS-Spoofed Ships

The Analysis

Strictly speaking, nothing that arrived on 20 June is a legal instrument, and I want to say that before I say anything else, because numbers this sharp invite people to read them as verdicts. They are observations. Windward's daily report for that day counted 32 vessels through the Strait of Hormuz, 17 inbound and 15 outbound, 11 of them dark, meaning no AIS broadcast at all. Inside the Gulf it counted 193 vessels showing GPS interference, the weight of it in the Kuwaiti and Emirati exclusive economic zones, and 120 vessels carrying a high-risk classification for smuggling. Two incidents of AIS identity splitting were detected, with real vessels' signals pushed to Null Island, zero degrees north and zero degrees east. At Kharg Island, 3 sanctioned VLCCs resumed loading, the first in roughly 14 days.

I am writing about this in a column on clauses and risk rather than leaving it to the operations people, because every one of those figures is a statement about evidence. Not about tonnage, not about rates, not about routing. About whether a position, an identity and a presence can be proved afterwards, in a room, months later, to someone who has no interest in being convinced. Cargo interests pay for proof. Insurance responds to records, charterparties are administered on records, sanctions files are judged on records, and a bill of lading is a record. On 20 June a meaningful share of the records being generated in that water were either absent or untrustworthy, and the interesting question is not what the ships were doing. It is what you would be able to show in eighteen months.

Here is the exception I would circle before anyone does anything with the 120. A commercial risk classification is not a legal designation. Being flagged as high-risk for smuggling by a private analytics provider carries no legal consequence on its own; being named on a sanctions list does, and the two are not the same population. I am going to return to that distinction more than once, because it is where I expect the expensive mistakes to be made, and the distance between a score and a designation is measured in what you can prove rather than in what you suspect. Strictly speaking, most of what follows is not advice about the Gulf. It is about the four documents you already have, and about how far these numbers say they can be trusted.

Start with eleven out of thirty two. That is 34.4 percent, call it a third of everything that moved through the strait that day, and I want to slow down on the inference people reach first. The instinct is to read a dark transit as a ship doing something it should not. Sometimes that is exactly what it is. A transponder can also be off for reasons that have nothing to do with concealment, including the judgment of a master who decides that being visible in a particular stretch of water is a hazard rather than a protection.

Which of the two applies on any given transit turns on flag state rules, on the conventions governing the equipment, and on what the master recorded at the time, and I am not going to recite provisions I have not verified in front of me. The point does not need them. Whatever the reason, once the signal stops the machine-generated record stops with it, and afterwards there is no independent account of where that ship was during the hours it was inside the strait.

Now hold that against the way these things are actually settled. A great many clauses in a charterparty, in a policy and in a contract of carriage switch on geography and time: was the vessel inside a defined area, at a defined moment, when the event happened. Not on what anyone believes happened. On where the ship was. If the record for that window does not exist, the party who has to establish the fact cannot establish it, and which party that is comes out of your contract and your policy, not out of the report. So the practical consequence of a dark transit is not suspicion. It is a gap, and gaps in a record are allocated by contract long before they are ever discovered.

The 193 is a different defect and, in my experience, a worse one. It is worth separating jamming from spoofing, because they fail differently. Jamming denies a receiver a position, which produces a hole in the track that anyone looking at it can see. Spoofing supplies a position, a coherent, plausible, continuous one that happens to be false, and it produces a track that looks like evidence and is not. The two Null Island events are the demonstration. Zero north, zero east is open ocean off West Africa; no vessel transiting the Gulf is there. A track that places one there is not degraded data, it is invented data, and the report's own description of identity splitting says the harder thing out loud: one hull's identity appearing in two places at once. Once an identifier can be separated from the hull carrying it, AIS is not identity evidence. It is a claim about identity made by a device.

Two incidents is two incidents. I will not compute a frequency from a sample of two, and neither should anyone else; a rate derived from two events is not a rate, it is arithmetic dressed as a rate. What two does establish is capability. The capacity to split an identity exists, it was observed on 20 June, and from the standpoint of anyone who has to prove which ship carried which cargo, capability is the relevant finding. It does not matter how often it happens. It matters that a record showing a clean, continuous, plausible track can no longer be treated as answering the identity question on its own.

Which brings me to the part cargo interests consistently underestimate, and it is the part with the longest fuse. Every screening obligation you have runs on vessel identity, and the identifier that matters is the IMO number, not the name on a screen and not a call sign picked up from a broadcast. If identity can be split, then screening the wrong entity is possible even when the process looks correct, and the defence you will need is not 'we screened' but 'we screened against a registry-sourced identifier, on this date, from this source, and here is the output'. Sanctions exposure is strict in most of the regimes I have watched applied, meaning a good-faith mistake is not automatically a defence, and the difference between a defensible file and an indefensible one is usually the provenance of the identifier rather than the diligence of the person doing the search.

So here is the boundary I would draw around the 120, and it is the sentence I would want on a memo. One hundred and twenty vessels classified as high-risk for smuggling is a commercial assessment made by a company that sells assessments. It is not a legal status, it is not a finding by any authority, and it is not a prohibition. What it can do is tell you where to look harder, which is a very different thing from telling you what is true. The failure mode I have seen is a commercial team treating a risk score as though it were a designation and declining cargo on that basis, and the opposite failure of a compliance team ignoring it entirely. The correct posture sits between them and it is procedural: screen against the official lists, keep the timestamped output, and treat the score as a trigger for deeper enquiry rather than as an answer.

The Kharg Island line is where I would most want people to slow down, precisely because it looks like news. Three sanctioned VLCCs resumed loading, the first in about 14 days, after a re-closure. That is a fact about three vessels on one day. It is not a rate, it is not a programme, and it is not a signal that anything has been relaxed. Let me do the arithmetic and then show you why it does not do what you want it to do. Assume, and this is my assumption for the sake of the calculation rather than anything in the report, that a VLCC of this class lifts roughly 2 million barrels.

Three of them is about 6 million barrels. Assume further, again purely for arithmetic, a price of 70 dollars a barrel, and three loadings represent on the order of 420 million dollars of crude. Now stop. You cannot divide that by 14 and call it a daily export rate, because the 14 days is the length of a gap, not the period over which the three loadings were spread. You cannot multiply it by 26 and call it an annual programme, because three ships is an event and not a schedule.

And you cannot read it as a change in legal position, because a sanctioned vessel loading cargo is a fact about that vessel, while whether it violates anything turns on the regime, the parties, the applicable exceptions and the authorisations in place. That last question is one for counsel with the actual facts, and I am not going to answer it from a summary.

Let me put a book around the evidence gap, with every assumption stated, because this is the calculation that tells you whether the paperwork is worth doing. Assume a trading house with 24 Gulf-loading voyages a year. Assume the dark-transit share observed on 20 June, 11 of 32 or 34.4 percent, holds roughly across the year, which is a strong assumption and I am flagging it as one: that gives about 8 voyages a year with no AIS record of the transit. Assume separately that 3 of the 24 show some form of position anomaly in the year, a second assumption, chosen to be conservative against 193 vessels reported affected in the Gulf on a single day. That is 11 voyages out of 24, call it 46 percent, where the position record is either missing or contested. Now assume average cargo value of 6 million dollars a voyage, and assume that on one of those voyages a claim turns on where the ship was and the record cannot settle it.

Run the money on that one voyage. Assume the insurer or the counterparty reserves the claim in full pending position evidence, and assume the evidentiary gap adds 90 days to resolution, which is a quarter and is the sort of delay I have seen when the only available account is crew recollection. At an assumed carrying cost of 8 percent a year, 90 days on 6 million dollars is about 118,000 dollars. Assume outside counsel and expert time on a contested position question runs 80,000 dollars, which is an assumption and not a quote. Call the cost of one evidentiary failure roughly 200,000 dollars. Against that, the work I am describing here is a clause amendment, a data-preservation obligation and a screening protocol, and it is done once and reused for years. I am not going to put a number on the legal time, because that is a commercial judgment and not mine. But the comparison is the one to make before the next fixture, not after the next incident.

There is a second calculation and it runs in the opposite direction, which is why people forget it. Twenty June is also a date, and dates matter in contract. From the publication of a report of this kind, the state of knowledge of anyone who reads it changes, and the change is documentable. Many force majeure and war-risk provisions, in one wording or another, turn on whether the event was foreseeable or known at the time of contracting, and on whether the party invoking the clause took reasonable steps. I am not reciting your clause, which I have not seen.

What I am saying is narrower: as of 20 June, 'nobody could have known about GPS interference in the Gulf' is a much harder sentence to sustain than it was on 19 June, and if you are contracting now, the fact that you knew and contracted anyway goes into the file. That cuts both ways. It strengthens a cargo interest arguing that a carrier should have planned for delay, and it weakens a cargo interest arguing surprise when its own cargo was on a dark ship.

Which is the point I want to leave on the contract side, and it is the one I would raise in a meeting before anyone starts talking about routing. The distinction that decides most of these disputes is not whether a risk existed. It is whether the risk was known, and by whom, at the moment the contract was made. Risks that exist but are unknown are one legal animal, and risks that are known and priced, or known and ignored, are another. Everything you do in the next quarter should be aimed at making your own knowledge legible afterwards: dated, sourced, written down, and attributable to a person.

So what I would actually check, document by document, because mixing them up is how things get missed. On the insurance side the questions are narrow, and they are all about words rather than about markets: what are your trading warranties, is the Gulf call inside or outside them, is there a listed-area mechanism in the policy and does it require a separate endorsement for each transit, and what notice is required and to whom. Have your broker answer those in writing and give you the page references. Not a summary of the cover, the words.

On the charterparty or the booking note, four provisions deserve a reading with a pen. The war risk clause, and specifically what counts as a triggering event and who may refuse. The deviation provision, and I would separate three questions that usually get collapsed into one: may the master deviate for safety, who bears the extra time and cost if he does, and does the insurance follow the vessel. Those are answered in different documents and sometimes in different contracts. The laytime and demurrage provisions, and whether time lost to a closure runs against you or is carved out, because the difference between 'time counts' and 'time does not count' is the entire dispute in a delay case. And the notice provisions, which are the cheapest thing to get right and the most expensive to get wrong, because a late notice can be fatal to a claim that is otherwise sound.

On the bill of lading and the booking, the identity question is what I would fix. Get the IMO number onto the shipping instructions and onto the bill as a matter of routine rather than as a reaction. Have the carrier or the charterer warrant the identity of the performing vessel and give you a right to reject a substitution that has not been screened. And if your paper carries any wording about known risks, assumed risks or excepted perils, read it now, in peace, rather than during a claim, because the words that decide these cases are usually the boring ones nobody argued about at signing.

On the sanctions file the discipline is provenance and time. Screen the IMO number against the official lists of every regime that touches your transaction, not one, and keep the output with its timestamp and the list version. Re-screen at the points that matter: at booking, before loading, before discharge and before payment, because a vessel can be designated between two of those and the file has to show when you checked. If a commercial risk score is what prompted the check, record that too, and record that the score was not the basis of the decision. That single sentence has ended more arguments than any other line I know of in a compliance file.

Then the evidence-preservation clause, which is the piece almost nobody has and everybody wishes they had. Ask for a contractual right to the vessel's position data, AIS or otherwise, and to the voyage data recorder if it becomes relevant, with a stated retention period, a named custodian and a stated format. Two years is a reasonable retention to ask for, and I would ask for it in the fixture rather than after the event, because after the event you are asking for a favour. Pair it with a counterparty obligation to notify you within a fixed number of days of any transit without AIS or any detected position anomaly, and set the number low: five business days is workable, thirty is not.

Ownership, then, by date, because a column that ends without dates is not much use to anyone. In the week of 22 June, have your legal or claims lead pull the four documents I have named, policy, charterparty or booking terms, bill of lading template and sanctions procedure, and mark each one against the questions above. That is a review, not a negotiation, and it should be finished inside ten business days with a written list of gaps. By 30 June, send the insurance questions to the broker and require written answers with page references. By 31 July, put the IMO number field and the identity warranty into the standard shipping instructions template and into the booking checklist, and make it a required field rather than an optional one.

By 30 September, have the evidence-preservation and anomaly-notification language drafted and tabled for the next round of fixture or service agreement renewals, with a target of getting it into at least half of the contracts you renew in the second half of the year. By 31 December, have the sanctions procedure rewritten around provenance and timestamps, and have it tested: run a retrospective re-screen of the last twelve months of voyages against the current lists and record how many would have been caught by the new procedure and how many would not. That test is the only honest measure of a screening protocol, and the number it produces is the one to report upward. Then hold all four dates on a quarterly review, because the regimes and the lists move far faster than contracts do.

The pitfalls, and they are mostly about presuming. Presuming that a dark transit is unlawful, when the reason is not established. Presuming that a continuous AIS track proves identity, when two incidents on one day show it can be fabricated. Presuming that a risk score is a designation, which will cost you a commercial relationship and gain you nothing legally. Presuming that three VLCCs at Kharg is a programme, when the report says only that loading resumed. Presuming that a closure of days amounts to frustration of the contract, when the clause and the duration decide that and a delay is not a discharge. And the quietest one: presuming that because the strait reopened, the evidentiary problem went away with it. Reopening moves ships. It does not repair the record of the fourteen days before it.

One counterfactual is worth stating, because it is the condition on which everything above becomes less urgent. If the interference profile drops away and the dark-transit share falls back toward the low single digits, this becomes a clause-hygiene exercise rather than a live exposure, and the right response is to finish the document review and stop there. If it persists or widens, the same work becomes a pricing and contracting question every quarter, and the notice and evidence clauses stop being boilerplate and start being the thing that decides claims. I do not know which of those two worlds we are in, and I am not going to guess from one day. What I can say without overreaching is that both worlds are served by the same preparation, and only one of them gives you time to do it later.

My brief here is the rule and its preconditions, and I am going to keep to it. Where the ships should go, what the tonnage does, which port takes the volume: that is not my column and I am not going to improvise it. What I will say is this. On 20 June a third of the transits through the most consequential strait in the world left no machine record of being there, nearly two hundred vessels in the same water reported a position that could not be trusted, and two of them were demonstrably somewhere they were not. That is not a shipping story. It is an evidence story, and evidence is the only part of a dispute you can prepare in advance. Strictly speaking, none of it means anyone has done anything wrong. All of it means that if something does go wrong, the side with the record wins, and the record has to exist before it is asked for.

↑ Back to the key points

— By Dr. Ingrid Voss